Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 29, 2025Open Access

YuE: Scaling Open Foundation Models for Long-Form Music Generation

View Full Paper
Ask AI
Bookmark
Share

Authors

RYRuibin YuanHLHanfeng LinSGShuyue Guo

Discussion

Loading...

Member takes

Overview

This system demonstrates improved song generation in music, indicating advances in foundation model capability.

Key Points

  • YuE generates up to five minutes of music while ensuring lyrical alignment and coherent musical structure.
  • The model employs multitask, multiphase pre-training that enhances learning across different tasks, achieving high musical quality.
  • Innovative techniques like track-decoupled next-token prediction enable YuE to effectively address complex music generation challenges.
  • Utility in music understanding tasks has been validated, with results from YuE matching or exceeding the state-of-the-art on the MARBLE benchmark.

Cite This Study

Yuan et al. (2025) studied this question.

synapsesocial.com/papers/68da58d1c1728099cfd10b96https://doi.org/10.48550/arxiv.2503.08638
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1ACE-Step: A Step Towards Music Generation Foundation Model2025
  2. 2SongComposer: A Large Language Model for Lyric and Melody Composition in Song Generation2024 · 4 citations
  3. 3InspireMusic: Integrating Super Resolution and Large Language Model for High-Fidelity Long-Form Music Generation2025
  4. 4Practical and Reproducible Symbolic Music Generation by Large Language Models with Structural Embeddings2024
  5. 5Ainur: Harmonizing Speed and Quality in Deep Music Generation Through Lyrics-Audio Embeddings2024 · 1 citations